In a recent paper, Galbreath, Mordacq & Weiler (2006) concluded that the kouprey Bos sauveli was not a natural species, but rather a feral animal derived from hybridization between banteng Bos javanicus and zebu Bos taurus indicus. Here, we analyze two mitochondrial genes (cytochrome b and subunit II of the cytochrome c oxidase) for all the seven species of the subtribe Bovina, including new sequences for several specimens of banteng, zebu and gaur of Cambodia. Our analyses indicate that mitochondrial sequences of Cambodian banteng are divergent from those of Javan banteng (mean difference =4.27%), but similar to those of kouprey (1.25%). We propose two conflicting hypotheses to interpret these results: (1) the Cambodian and Indonesian banteng belong to two distinct species, and the kouprey derived from Cambodian banteng; (2) all subspecies of banteng belong to Bo. javanicus, but the mitochondrial genome of kouprey was transferred by natural hybridization into the ancestor of Cambodian banteng. Morphological, ecological and ethological characteristics of banteng and kouprey rather support the second hypothesis. However, we need to sequence nuclear markers, and to analyze banteng from Lao, Myanmar, Thailand and Vietnam, to give a definitive conclusion on the taxonomic status of banteng and kouprey.